NUMERICAL SOLUTION

Chapter 6 OF ORDINARY
DIFFERENTIAL EQUATIONS

6.1 INTRODUCTION

A
physical situation concerned with the rate of change of one quantity
with respect to another gives rise to a differential equation.
Consider the first order ordinary differential equation
dy
= f (x, y) (1)
dx
with the initial condition
y(x0) = y0 (2)
Many analytical techniques exist for solving such equations, but these
methods can be applied to solve only a selected class of differential equations.
However, a majority of differential equations appearing in physical problems
cannot be solved analytically. Thus it becomes imperative to discuss their
solution by numerical methods.
In numerical methods, we do not proceed in the hope of finding a
relation between variables but we find the numerical values of the dependent
variable for certain values of independent variable.
It must be noted that even the differential equations which are solvable by
analytical methods can be solved numerically as well.

479
480 COMPUTER-BASED NUMERICAL AND STATISTICAL TECHNIQUES

6.2 INITIAL-VALUE AND BOUNDARY-VALUE PROBLEMS

Problems in which all the conditions are specified at the initial point only are
called initial-value problems. For example, the problem given by eqns. (1)
and (2) is an initial value problem.
Problems involving second and higher order differential equations, in which
the conditions at two or more points are specified, are called boundary-value
problems.
To obtain a unique solution of nth order ordinary differential equation, it is
necessary to specify n values of the dependent variable and/or its derivative at
specific values of independent variable.

6.3 SINGLE STEP AND MULTI-STEP METHODS

The numerical solutions are obtained step-by-step through a series of equal

intervals in the independent variable so that as soon as the solution y has been
obtained at x = xi , the next step consists of evaluating yi+1 at x = xi+1. The
methods which require only the numerical value yi in order to compute the
next value yi+1 for solving eqn. (1) given above are termed as single step
methods.
The methods which require not only the numerical value yi but also at
least one of the past values yi–1, yi–2, ...... are termed as multi-step methods.

6.4 COMPARISON OF SINGLE-STEP AND MULTI-STEP METHODS

The single step method has obvious advantages over the multi-step methods
that use several past values (yn, yn–1, ......, yn–p) and that require initial values
(y1, y2, ......, yn) that have to be calculated by another method.
The major disadvantage of single-step methods is that they use many more
evaluations of the derivative to attain the same degree of accuracy compared
with the multi-step methods.

6.5 NUMERICAL METHODS OF SOLUTION OF O.D.E.

In this chapter we will discuss various numerical methods of solving ordinary

differential equations.
We know that these methods will yield the solution in one of the two forms:
(a) A series for y in terms of powers of x from which the value of y can be
obtained by direct substitution.
 NUMERICAL SOLUTION OF ORDINARY DIFFERENTIAL EQUATIONS 481

(b) A set of tabulated values of x and y.

Picard’s method and Taylor’s method belong to class (a) while those of
Euler’s, Runge-Kutta, Adams-Bashforth, Milne’s, etc. belong to class (b).
Methods which belong to class (b) are called step-by-step methods or
marching methods because the values of y are computed by short steps ahead
for equal intervals of the independent variable.
In Euler’s and Runge-Kutta methods, the interval range h should be kept
small, hence they can be applied for tabulating y only over a limited range.
To get functional values over a wider range, the Adams-Bashforth, Milne,
Adams-Moulton, etc. methods may be used since they use finite differences
and require starting values, usually obtained by Taylor’s series or Runge-Kutta
methods.

6.6 PICARD’S METHOD OF SUCCESSIVE APPROXIMATIONS

Picard was a distinguished Professor of Mathematics at the university of Paris,

France. He was famous for his research on the Theory of Functions.
Consider the differential equation
dy
= f (x, y); y(x0) = y0 (3)
dx
Integrating eqn. (3) between the limits x0 and x and the corresponding
limits y0 and y, we get

z z
y

y0
dy =
x

x0
f ( x, y) dx

⇒ z
y – y0 =
x

x0
f ( x, y) dx

or, y = y0 + z x

x0
f ( x, y) dx (4)

In equation (4), the unknown function y appears under the integral sign.
This type of equation is called integral equation.
This equation can be solved by the method of successive approximations or
iterations.
To obtain the first approximation, we replace y by y0 in the R.H.S. of eqn. (4).
Now, the first approximation is

y(1) = y0 + z x0
x
f ( x, y0 ) dx

The integrand is a function of x alone and can be integrated.